A systematic review and meta-analysis of interventions to increase stroke thrombolysis

Although the efficacy of tissue plasminogen activator (tPA) for acute ischemic stroke is well established [1, 2], rates of tPA use remain low both nationally and globally. Estimates of tPA treatment rates in the United States are consistently less than 5% [3, 4]. ,and similar rates are found in the United Kingdom [5]. Given that stroke thrombolysis is either a cost-saving or highly cost-effective therapy, interventions to increase thrombolysis are likely to be sound societal investments [6‐9].

Regional rates of tPA use range from 0 to 9.3%, suggesting both that regional factors may influence treatment rates and that considerable opportunity for improvement exists [10]. While the factors that underlie this variation are not well established, access to stroke care explains only a fraction of these differences [11]. Other factors potentially implicated in low treatment rates include poor symptom recognition by the public [12, 13]; failure of timely transport to emergency rooms [14, 15]; and underdeveloped emergency department (ED) systems of care [16, 17].

Interventions aimed at increasing tPA use have been developed to address these factors. These interventions use a variety of approaches in various settings to increase treatment rates. For clinicians, advocates, and policy-makers seeking to increase tPA treatment rates, it is important to both understand what interventions exist and to understand the relative effectiveness of these interventions. Addressing this knowledge gap may help hospitals and communities increase their acute stroke treatment rates.

To inform future initiatives to increase tPA use, we performed a systematic review and meta-analysis of existing interventions to increase tPA rates for acute ischemic stroke. Our primary goals were to: 1) quantify and describe intervention approaches and settings, 2) determine the overall effectiveness of these interventions, and 3) explore the relative effectiveness of particular intervention approaches.

In this systematic review, we found numerous interventions that used a variety of approaches to increase tPA treatment rates in different settings. The intervention studies were reasonably designed to address the effectiveness of the tested interventions. The magnitude of benefit of the average intervention was substantial, and a trend toward benefit was observed across all intervention approaches studied. The overall summary risk ratio of 1.80 means that, if a region has a 5% treatment rate, this rate would increase to 9% with the average intervention. However, whether this effect size is an accurate representation of the real world effect of all attempted interventions is less clear.

Our review uniquely contributes to the literature by comprehensively exploring interventions that have been utilized to increase thrombolysis use. As such, our analysis attempts to inform the question for a region or hospital about how best to increase tPA treatment rates. We did not find strong evidence that one intervention approach was associated with substantial differences in treatment rates compared to other intervention approaches as the point estimates were similar across intervention type. Given the small number of included studies and their heterogeneity, the data do not support strong conclusions about which intervention approach is best. Because interventions involving EMS were most common, the statistical evidence is most robust for EMS-based interventions. Therefore, of all the interventions identified, it is most likely that EMS-based interventions are efficacious. However, we cannot assume that the benefit of EMS-based interventions is of greater magnitude than the benefit of other intervention approaches.

Rather than selecting an intervention approach, a region seeking to maximize improvement in its treatment rates may wish to first determine its largest treatment barriers and then accordingly select a strategy, as it appears that most intervention approaches have relatively similar overall effects. For regions with long pre-hospital delay – focus on public education; for regions with prolonged time from EMS activation to ED arrival – focus on EMS; for regions without access to adequate tPA support – focus on telemedicine, and for regions with long door to treatment times – focus on hospital quality improvement. As many of the interventions suggest, focusing on more than one barrier at a time is eminently feasible.

Once regional barriers are identified, a number of excellent exemplar interventions exist. In their Stockholm-based study, Berglund et al. found that simply increasing the EMS priority level of stroke patients from two to one resulted in a 14% absolute increase in treatment rates (p < 0.001) [24]. Quain et al. implemented a Pre-hospital Acute Stroke Triage (PAST) protocol, which included a pre-hospital stroke assessment tool for EMS, an established protocol for hospital bypass, and pre-hospital notification. The rate of thrombolysis in that study increased from 4.7% pre-intervention to 21.4% post-intervention (p < 0.001) [25]. In their Bavaria-based study, Müller-Barna et al. studied the effectiveness of the TeleMedical Project for integrative Stroke Care (TEMPiS), an effort which sought to improve stroke outcomes in rural areas. This initiative consisted of 24/7 virtual availability of a vascular neurologist at 15 regional hospitals [26]. Rates of tPA administration increased from 0.4% in the year before the intervention (2002) to 15.5% in the intervention’s tenth year (2012). Finally, the TLL Temple Foundation Stroke Project used a public education approach to improve acute stroke recognition and treatment by employing a community-based mass media campaign, community volunteer training, and the establishment of multidisciplinary teams in hospitals to develop ED protocols [27]. In this study, the percentage of ischemic stroke patients treated with IV tPA was 8.65% in the intervention group compared to 0.86% in the control group. In addition to the interventions identified in our review, since completion of the review, the results of additional studies have suggested that mobile stroke units are effective in increasing tPA treatment rates in selected regions [28‐30].

Our study has several limitations. The pattern of missing data suggests publication bias which may result in an overestimate of the summary effect size. If all interventions were published, we would anticipate that small studies with relatively large positive effects would have similar representation to small studies with relatively small effects. However, we found that the smallest studies in our sample universally had large positive effects, suggesting publication bias. On the other hand, there was no evidence that effect sizes varied by study design, as similar effect sizes were seen in the strongest study designs (cluster RCTs) and the rest of the reported studies. In addition, the statistics for heterogeneity among studies is high given our analysis of varied interventions. Interpretation of the pooled effect sizes should be made with caution due to the heterogeneity among the studies. We have followed the Cochrane handbook for systematic reviews of interventions by reporting our random effects analysis as our primary analysis as it better accounts for heterogeneity.

Thirteen of the 25 studies in our review were pre-post design. Pre-post studies can be biased as data may be differently collected (e.g. more systematically) after an intervention than before an intervention.

Furthermore, many governmental, regional, and organizational interventions to improve access to thrombolysis are not published as studies or trials in journal articles. Our meta-analysis fails to capture these interventions. In addition, because our literature review was completed in 2015, our analysis includes limited studies about the effectiveness of mobile stroke units in increasing thrombolysis rates [31, 32].

Given the magnitude of the summary effect size, the number of studies included, and the consistency of the positive effect, it is likely that interventions have a net positive effect of a magnitude that is clinically and societally meaningful. We did not prospectively register our meta-analysis with a register given that, at the time we undertook this study, meta-analysis registration was not widely performed.

In summary, a considerable body of evidence exists supports a variety of interventions to increase tPA use. On the whole, these interventions appear to have considerable effectiveness and our findings support the pursuit of future interventions to increase the rate of tPA administration.